Peltier tec question

[polishedball]

Hoping someone can help me with this.

What would a unconnected tec do if sandwiched along side a working unit. Not stacked. Will it act as a heat pump, possibly stealing the cold from the plate above or transfer heat to the cold plate side? Basically just thinking about adding a second I case it is needed for more cooling, that way a tear down wouldn't be needed. Just didn't know how it acts if not powered. Unsure as to if this could cause problems.

Thoughts?

Thanks

[The_Doctor]

They conduct heat too well to allow good use of the working one. It would be more efficient to underpower both of them. This sill still be under best efficiency but nowhere near as bad. Also, the spread of thermal differences will be pretty similar most times if both TEC's are driven equally, which will reduce the chances of thermally induced alignment problems.

Also, calculating heat loads and anything else with one on, one off, will be horrible. Using two in series or parallel (series electrically, not thermally) amounts to calculating for one TEC whose area equals the two used together. Some very strong TEC's are inefficient anyway, so you might gain from using two more modest ones if you have the area and the capacity to remove their heat easily. Don't add too much heat pumping capability because a big pump takes more power to pump a small amount of heat than a smaller one that can do it. Just figure out worst case expected, and plan for that.

[polishedball]

I have a dozen or more 12v 9A pelts. So the issue is for me to add a second I'd also need another 12v 12a supply, wanted to save the space and $$ if possible. But sounds like it needs to be powered if present.

[The_Doctor]

The cheap, common 40mm square ones all over eBay, Amazon, etc? Like TEC1-12709?
If so, their spec goes (roughly) like this:
Rated 12V, 9A.
Actually means they'll draw 9A at 12V, consuming 108W to be got rid of plus whatever heat gets pumped, while operating at best efficiency.
Meaning that at Qmax (of 82W, these are often incorrectly stated as 90 or more) they'll need 15.4V. The Q will be about 20~30% lower than max on 12V.

If you've planned space, then it may be better to get a stronger 40mm square TEC, but start by figuring out your total load, then double it (before adding whatever the TEC itself will consume when pumping it). If your thermal losses are low you might get by with less than double, but quantifying that is very hard to do. Far easier to double it and run the TEC at less power than best efficiency point wants, because it will still be close to best efficiency.

EDIT:
That load-doubling estimate is based on small loads with large complex areas in small enclosures where thermal isolation is difficult. You'd need to look at Melcor or Marlow design guides as an example for other situations, but I'm assuming you're cooling a small diode rig.

[logsquared]

I believe the off state thermal conductivity is very low. Around 3-4 W/ mK if recall. Compare that to aluminum at around 200 W/mK. I dont think you would loose much through a passive element in parallel.

I have found that most of the TEC's i have used have varied a few thousandths from piece to piece in thickness. This caused me grief in the past trying to put them next to each other.

You may find that running them in series (electrical) will give you the cooling power you need due to increased efficiency.

[The_Doctor]

I believe the off state thermal conductivity is very low. Around 3-4 W/ mK if recall. Compare that to aluminum at around 200 W/mK. I dont think you would loose much through a passive element in parallel.

I have found that most of the TEC's i have used have varied a few thousandths from piece to piece in thickness. This caused me grief in the past trying to put them next to each other.

You may find that running them in series (electrical) will give you the cooling power you need due to increased efficiency.

Thermal conductivity is kept as low as possible to increase cooling capability, but with an area of 800 square millimeters it's still going to be by far the biggest parasitic heat load in the system! A quick test is to touch an unpowered, unmounted 40mm TEC at room temperature with the back of two fingers. It will feel cool, or cold, and it will stay that way for several seconds. That can't happen unless it is conductive enough to keep pulling heat out of your fingers for a while. To get a stronger effect you'd have to do it with a chunk of metal. Compared to most materials, a TEC is very conductive.

Also, for best efficiency each TEC must be considered in its own right, so the system will not be more efficient in series, or parallel, the choice there is purely a convenience basis for PSU design. Series might have a (very) tiny advantage in using slightly thinner wires with less heat produced, but the amounts will be tiny, far smaller than most of the other parasitic losses, especially if one TEC is unpowered.

[logsquared]

Also, for best efficiency each TEC must be considered in its own right, so the system will not be more efficient in series, or parallel, the choice there is purely a convenience basis for PSU design. Series might have a (very) tiny advantage in using slightly thinner wires with less heat produced, but the amounts will be tiny, far smaller than most of the other parasitic losses, especially if one TEC is unpowered.

I meant more efficient than running one element harder. Given the 12V supply mentioned wouldn't drive them in parallel I suggested series. Should have been more specific.

If it were me I would put both elements in now paying attention to the flatness of the metal parts and thickness of the TEC's. First I would try them series. Next, one element, if series didn't work. Third, get a bigger supply and go parallel

[andy_con]

operates to -60 if that helps?

http://www.ebay.co.uk/itm/Gigantic-6...item48357d2b88

[The_Doctor]

operates to -60 if that helps?

http://www.ebay.co.uk/itm/Gigantic-6...item48357d2b88

I read recently that those were horribly inefficient, but that might just be because it's a bitch trying to get that many watts through so small an area without extremely good thermal interfacing..

Logsquared, point taken, if two are pushed to do one's work, the deviation from best efficiency is on the steeper bit of 'bell curve which isn't really a bell curve at all' so it would be more efficient, (and would be so in electrical series or parallel). Regarding PSU, I assume that volts has more headroom available than amps, or vice versa. If you optimised for the better headroom you might scrape by with the same PSU, especially if you can improve the thermal isolation of mounting hardware, etc. Series would be great if that PSU can be tweaked to do 15V or more.

[lasersbee]

Hoping someone can help me with this.

What would a unconnected tec do if sandwiched along side a working unit. Not stacked. Will it act as a heat pump, possibly stealing the cold from the plate above or transfer heat to the cold plate side? Basically just thinking about adding a second I case it is needed for more cooling, that way a tear down wouldn't be needed. Just didn't know how it acts if not powered. Unsure as to if this could cause problems.

Thoughts?

Thanks

I've found some useful info on cascaded/stacked TECs here...

http://www.tecpeltier.com/viewtopic.php?f=4&t=19

Hope it helps.


Jerry